Precise feeding control system based on machine learning

The invention relates to the field of poultry farming, in particular to an accurate feeding control system based on machine learning, and the system comprises a detection module which detects a plurality of temperature values of a feed in a trough at any time through a temperature sensor group and detects a plurality of humidity values of the feed contained in the trough through a humidity sensor group; the central control module obtains a temperature average value of the feed according to the plurality of temperature values and compares the temperature average value with a preset standard temperature, and when the temperature average value is not equal to the preset standard temperature, a heating device or a cooling device below a circulating water pipe connected with a constant-temperature water tank at the bottom of the trough is started; and when the humidity average value is larger than the preset standard humidity, the opening and closing angles of an inlet and an outlet of a constant-temperature water